Princeton and Brookhaven Lab scientists collaborate to find sources of quantum information loss at the atomic level.

Scientists and engineers studying superconducting quantum bits (qubits), a top quantum computing platform that relies on the frictionless flow of paired electrons, have discovered clues to the micro-sources of qubit information loss. This is one of the main obstacles to realizing quantum computers that can string together millions of qubits to perform complex computations. This fault-tolerant, large-scale systems can simulate complex drug-development molecules and accelerate the discovery of clean energy materials. They also have the potential to perform other tasks that are impossible or too time-consuming (millions of Years) for today’s supercomputers.

Due to atomic-scale defects, it is still challenging to understand the nature of qubit-information loss. This gap between qubit performance and material properties was bridged by the team using state–of–the–art characterization capabilities at both the Center for Functional Nanomaterials and National Synchrotron Light Source II, both U.S. Department of Energy Office of Science User Facilities at Brookhaven National Laboratory. The team discovered structural and surface chemistry problems in superconducting Niobium qubits, which may have led to the loss.

“Superconducting qubits offer a promising platform for quantum computing because we can engineer and make their properties using the same tools that make regular computers,” Anjali Premkumar said, a fourth-year graduate student at Princeton University’s Houck Lab and the first author of the Communications Materials paper detailing the research. They have a shorter coherence time than other platforms.

They can only keep information for a short time before it is lost. Although coherence times for single qubits have improved from microseconds to milliseconds, they decrease significantly when multiple qubits are interconnected.

Premkumar said, “Qubit coherence can be limited by the quality superconductors as well as the oxides that will invariably grow on them when the metal comes into contact with oxygen in the atmosphere.” “But, as qubit engineering, we have yet to characterize our material in great detail. We collaborated with material experts to examine the structure and chemistry of our materials using sophisticated tools for the first time.

This collaboration was a “prequel” to the Co-design Center for Quantum Advantage, C 2QA, one of five National Quantum Information Science Centers established in 2020 to support the National Quantum Initiative. Brookhaven Lab’s C 2QA is a collaboration of hardware and software engineers, physicists and materials scientists, theorists, and other experts from national labs, universities, and industry to solve performance problems with quantum hardware. The C 2QA team focuses on materials, devices, and software co-design. They seek to understand and control material properties to extend coherence times and design devices that generate more robust qubits.

The team created thin films of niobium through three different sputtering methods. Sputtering is where energetic particles are fired at a target that contains the desired material. Atoms are then ejected from this target and land on a nearby substrate. The standard form of sputtering used by the Houck Lab is direct current. Angstrom Engineering uses a type of sputtering called high-power impulse magnetron (or HiPIMS). This involves striking the target with short bursts o high-voltage energy. Angstrom performed two kinds of HiPIMS: the normal and the optimized power versions.

Premkumar, a Princeton graduate, made “transmon,” qubit devices using the three sputtered films. He then placed them in a refrigerator for dilution. The fridge can drop near zero (minus 459.67 F), making qubits superconducting. These devices allow superconducting pairs to tunnel through an insulating barrier made of aluminum oxide (Josephson Junction). This is sandwiched between superconducting layers of aluminum, which are coupled with capacitor pads of sapphire niobium. The qubit state changes as the electron pairs move from one side to the other. Transmon qubits were co-invented jointly by Andrew Houck, Lab principal investigator, and C 2QA Director Andrew Houck; they are an essential type of superconducting qubits. They are susceptible to fluctuations in the environment’s electric and magnetic fields. These fluctuations can cause qubit information losses.

Premkumar measured energy relaxation time for each type of device. This quantity is related to the qubit state’s robustness.

“The energy relaxation period corresponds to how long the qubit remains in the first excited condition and encodes the information before it decays into the ground state and loses the information,” said Ignace Jarrige, a physicist with NSLS-II, and is now a quantum research scientist at Amazon.

Each device had different relaxation times. The team used microscopy and spectroscopy at NSLS-II and CFN to understand the differences.

The NSLS-II beamline researchers determined the oxidation state of niobium using x-ray photoemission spectroscopy. This included soft x-rays from the In situ and Operando Soft X-ray Spectroscopy beamline and hard x-rays from the Spectroscopy Soft & Tender (SST-2) beeline. These spectroscopy studies revealed suboxides located between the metal layer and the oxide layer, which contained a lower amount of oxygen than niobium.

Jarrige explained, “we needed the high energy resolution of NSLS-II to distinguish the five different oxidation states of niobium as well as both hard and soft x-rays. These have different energy levels to profile these states according to depth.” “Photoelectrons produced by soft xrays cannot escape the surface for more than a few nanometers, while those created by hard xrays can escape the film’s depths.”

In the NSLS II Soft Inelastic X-ray Scattering beamline (SIX), the team detected spots with missing oxygen atoms using resonant x-ray scattering. These oxygen vacancies can absorb energy from qubits and are called defects.

The CFN team visualized film morphology with transmission electron microscopy. They also used atomic force microscopy to characterize the chemical makeup of the film surface using electron energy-loss spectroscopy.

Coauthor Sooyeon Hwang is a CFN Electron Microscopy Group staff scientist. She explained that the microscope images showed grains, which are pieces of individual crystals with atoms in the same orientation. They were sized smaller or larger depending on the sputtering method. “The smaller grains indicate more grain boundaries or interfaces between different crystal orientations. The electron energy-loss spectrum showed that one film contained oxides on its surface and oxygen within the film, with oxygen dissolved into the grain borders.

The experimental results at the CFN-II and NSLS-II showed correlations between qubit relaxation time and the number and breadth of grain boundaries, and the concentration of suboxides close to the surface.

Premkumar stated that grain boundaries are defects that can dissipate energy. Too many can impact electron transport and the ability of qubits to perform computations. “Oxide quality” is another critical parameter. Suboxides can be dangerous because electrons cannot happily pair together.

The team will continue to work together to understand qubit coherence using C 2QA. One research direction is to determine if relaxation times can be improved by optimizing fabrication to produce films with larger grain sizes (i.e., minimal grain boundaries) and a single oxide state. They will also explore other superconductors, such as tantalum, which has a higher chemical uniformity.

Premkumar stated, “From this study, we now have the blueprint for scientists who make qubits as well as scientists who characterize them to collaborate to understand microscopic mechanisms that limit qubit performance.” “We hope that other groups will use our collaborative approach to propel the field of superconducting qubits forward.”